Hard disk drives are well known and commonly used storage devices. Accordingly, although the present invention is not limited to use with hard disk drives, they will be used to illustrate the background of the invention and problems associated with the prior art.
In the field of digital data storage systems, most commercial computers employ Winchester-type disk drives to store program software and data. The predominant commercial hard disk drive standard is referred to as the 51/4" disk drive. Recent advances in hard disk drive technology have resulted in a new disk drive standard, commonly referred to as the 31/2" disk drive, which can store considerably more data in a substantially smaller configuration as compared to earlier 51/4" disk drives. The storage capacities of these smaller 31/2" disk drives continue to increase, with the 2112 series units made by Micropolis, the Assignee of the present invention, now having a storage capacity in excess of one gigabyte (equivalent to 1,000 megabytes). A "byte" of digital information includes eight binary digits or "bits" of information, and one gigabyte represents one billion bytes or eight billion (8,000,000,000) bits of digital information.
The data storage capacity of a particular disk drive depends on the user's anticipated storage requirements. However, it is often difficult to add data storage capability to a computer system once it has been configured. This difficulty is usually more dominant with personal computer type systems, but can present itself with work stations and larger systems.
External disk drive storage systems have provided the capability for adding internal or external disk drives to replace or augment the internal disk drive of a personal computer or work station. Adding such components may result in a number of problems related to power supply. In most systems, a single power supply is coupled to a number of individual peripheral components, including disk drives. The capacity of the system power supply is predetermined, and is based on the expected load that the system is anticipated to draw. Increasing the number of peripheral devices, or the capacity of peripheral devices such as disk drives, results in a requirement for additional power which may have been unanticipated. As a result, it may be necessary to replace the power supply once it has been determined that it is necessary to expand the system capabilities.
When the power supply capabilities are expanded, an attempt is made to determine whether the new power supply will accommodate the future power supply requirements of the system. Future expansion beyond the anticipated load necessitates replacing the power supply again.
In addition to the issues discussed above with respect to changes in data storage capacity and power supply capability, disk drive failure presents still another challenge for computer users. The tremendous storage capacity of these state-of-the art disk drives allows the computer user to store vast amounts of data on a single disk drive. Any failure of the disk drive or its supportive components, such as a power supply or cooling fan, can compromise the integrity of the stored data and cause substantial delays associated with repairing the unit and regenerating the data.
In order to minimize catastrophic loss of data, numerous software products and hardware storage devices have been developed to "back-up" the data residing on the hard disk drive of a computer. Data is typically backed up on a daily or weekly basis which requires substantial time to transfer the data from the hard disk drive to an external storage device. Incremental backups can reduce backup time by transferring only those files that had been modified since the previous backup. In the event that a hard disk drive failure occurs, the computer must be taken off-line, the hard disk drive replaced or repaired, and the backed up data restored to the new or repaired disk drive from the external backup device, thereby resulting in appreciable computer down-time.
Local area networks (LANs) have now become the preferred way of sharing data amongst a plurality of remotely located computers. Typical LAN configurations include one or more file servers that store large volumes of commonly shared data and distribute this data to remote users over communication lines. An individual file server is typically comprised of three hard disk drive units which store, among other data, the network operating system software, user software, and user data.
The cost of individual file servers often precludes acquisition of additional backup file servers, which often remain idle until needed, to replace a primary file server should the primary server fail. When a backup file server is available, data must be transferred to the backup server from an external storage device, typically a high density magnetic tape storage system. All data generated subsequent to the last backup, however, is usually lost.
Computer and network down-time can severely impact user productivity and often results in the loss of vital data. To minimize data storage system down-time, a method of simultaneously storing data on a plurality of disk drive units, often termed as "mirroring" data, has been developed to prevent the destruction of data and reduce system down-time resulting from hard disk drive failures. These "fault tolerant" storage systems, however, are often prohibitively expensive, occupy significant floor space, require special environmental cooling control and, most significantly, require the system to be taken off-line to repair one or more of the defective disk drive units.
Fault tolerant systems may also be vulnerable where the plurality of included disk drive units share a common power supply and/or disk controller circuitry which can compromise system integrity should a failure in these critical components occur. Furthermore, many multi-drive storage systems are housed in a chassis having a predetermined configuration which provides space for only a limited number of disk drive units, typically three such units making expansion of data storage resources quite expensive. In many cases, one of the included disk drive units is wholly or partially dedicated for storing parity data which provides information regarding the data stored on the remaining disk drive units.
Moreover, such fault tolerant systems are configured such that the plurality of disk drive units are connected in parallel, with no capability to partition or segment the system into two or more independent sub-systems. Such systems may not provide the flexibility or capability to interconnect the plurality of disk drives in series as to mimic a single disk drive, with the total storage capacity being equal to the cumulative capacities of the individual disk drives.
Prior fault tolerant system designs, similar to those disclosed in U.S. Pat. No. 4,870,643 (Bultman et al.) and U.S. Pat. No. 4,761,785 (Clark et al.), suffer from many of the deficiencies discussed hereinabove. Neither of the references cited above incorporate the advantages of expandability and flexibility inherent in the present invention.